Everything you need to launch an intelligence business
Your seats are about to get smarter. This is the resource that gets you there: what each team owns, what to decide, and what to send back before your launch coordination meeting.
How to use this guide
Launching this is a short project with five or six owners across your business. Each chapter is written for the team that owns that part, and tagged so you can filter this page and forward just what applies.
Nothing is filled in here. Every question and every field lives in one workbook, which you complete and return before the launch coordination meeting.
Launch Configuration Workbook
Eight tabs, one per team. The launch plan, contacts, branding, SKUs, platform configuration, legal, and your internal user list.
The link opens view only. Choose File, then Make a copy, to create your own. Complete it, then share your copy with your Tresic launch contact five business days before the coordination meeting.
Who this is for
Naming, packaging, price
Three decisions that gate everything else.
SKUs and quoting
Copy-and-paste item names and descriptions.
What arrives each month
Sample usage file and invoice, so you can map it before go-live.
Contracts and consent
Your customer agreement will need an addendum.
Connectivity and access
API keys, DNS, email routing, recording.
Launch to your base
A message library, email templates, and the asset list.
What happens next
Trained, branded, and in market within thirty to sixty days of contract, where your platform is one we already connect to. Three lanes run in parallel from the day you sign.
Read and return
Circulate the chapters, complete the workbook.
Coordination meeting
Decisions confirmed, credentials issued, dates set.
Build in parallel
Platform live, items quoted, teams trained.
Launch and run
Base announced, first quotes carrying the line.
What you are launching
One licensed seat that attaches to a seat you already sell. It runs on the phones your customers already have. Nothing to install, nothing to migrate, and no second system for the end user to learn.
Every seat carries all three capabilities below, plus 300 processing minutes per seat per month, pooled across the account, and API access included in the seat price.
A written summary of the conversation, and every commitment made in it captured as a task you can assign and set a date on. Searchable history across everything the user has said and heard. The summary is table stakes now. The tasks are the product.
- Summary after every conversation, same business day
- Commitments captured as tasks, ready to assign
- Searchable conversation history
- Scoped to the individual and their extension
- Nothing during the call. No prompting, no whisper, no screen pop.
- No manager view across a team in the base seat
- No scoring, grading, or coaching
- Manager visibility across a team's conversations and open follow-ups in one place
Two triggers run on every conversation and surface to the people the business nominates, within minutes of hang-up. The base triggers are customer frustration and customer threats, meaning a caller signalling they are ready to leave or escalate.
- Two triggers, running on every conversation
- Delivered within minutes of hang-up
- Recipients nominated by the business
- One organization, one alert group
- Alerts a person. Never routes, holds, or acts on a call.
- No custom triggers in the base seat
- No separate routing by team, department, or site
- Triggers built around their business, routed separately for each team, department, or location
What the account is hearing across every conversation, answered through a set of prepared questions. The pulse of the business without anyone listening to a single call.
- Prepared questions across the account
- Updated same business day
- Account treated as one pool
- Access granted to whoever the business chooses
- No free-text questions at any tier. A typed question is recorded for review, not answered.
- No breakdown by team
- No location dimension on the dashboard
- No individual performance reporting, by design
- The same prepared questions, answered across the whole organization and the full retained history
It surfaces and notifies. It does not route a call, act on a call, or make a decision, and nothing happens while the conversation is still going. Everything begins after hang-up.
Platform requirements and considerations
Four things to confirm before a rep quotes. These belong in your qualification script rather than being discovered at provisioning.
| Confirm | Why |
|---|---|
| Recording is enabled | Recording must be on for every user or line that will carry a seat. No recording, no product. This is the most common reason a deal stalls at activation. |
| The platform is reachable | We retrieve the recording and the call detail record as soon as a conversation completes. Chapter 09 covers what your platform team confirms once, for all accounts. |
| The account lives on the phone | The value is in what people say. An account whose staff barely use the phone is a poor fit regardless of headcount. |
| Minutes are understood | Total monthly minutes divided by seats tells you what to package. Chapter 04 works this through. |
If the account already pays for transcription
Some customers already have transcription switched on at the platform level. Most do not. Where they do, they are about to pay for it twice, and the usual answer is to turn the platform transcription off.
- There is no connector that imports existing transcripts. We work from the recording. Transcripts already sitting on the platform stay where they are.
- Retrieval is the actual upgrade. Accounts that have been recording and transcribing for years typically cannot find anything in it. For them the pitch is not transcription. It is being able to find and act on what was said.
When a customer is configured, an industry can be selected, and the topic taxonomy behind it sharpens what the product looks for. Two industries are tuned to depth today and the rest carry baseline taxonomies, so treat this as a configuration step rather than a reason to chase a vertical. Healthcare has a specific answer in Chapter 07.
Naming and branding
This is your product. Your name, your brand, your customer. Our name appears nowhere your customers can see it. Three decisions, and all three are needed before we configure anything.
The product name
What appears on your quote, your site, and your invoice.
The access URL
A complete address of your choosing.
Notification email
Our infrastructure or yours. Tell us where to point.
The product name
You choose it. Intelligence Cloud is available to use as-is if you want it, and several partners do. If you want your own name, nothing about the configuration changes.
Whatever you pick has to work in four places at once, which is the real constraint:
| Test | Question |
|---|---|
| Catalog | Does it fit on a quote line without a subtitle? |
| URL | Does it read as yours when it sits in front of your domain? |
| Sender | Does a notification from that name look legitimate to someone who has never heard of it? |
| Rep | Can a rep say what it is in one sentence without using the word platform? |
The access URL
Your users need somewhere to sign in. You choose the full address and it is yours entirely.
| Option | Example | Setup |
|---|---|---|
| Your domain | intelligence.yourcompany.com or any subdomain you own | You add a DNS record. We enable the address. |
| A subdomain of ours | yourcompany.intelligence.cloud | We configure it. Nothing for you to do. |
Keep the product name recognizable in the address. Your support team and your customers will both need to remember it.
Notification email
The product sends invitations, alerts, and digests. Those need a from-address, and you choose whose infrastructure they leave from.
| Option | What the recipient sees | What we need |
|---|---|---|
| We send them | A sender on our infrastructure, carrying your display name | The display name. Nothing else. |
| You send them | A sender on your own domain, delivered through your mail service | SMTP host, port, sender address, and credentials. Any SMTP-enabled service works, including whatever already sends your platform notifications. |
If mail should leave from your domain, give us the address and the credentials to relay through, and we point notifications there.
The word that will cost you: conversational
Two phrases in this market differ by three letters and mean opposite things.
| Phrase | What the market hears | Category |
|---|---|---|
| Conversational AI or intelligence | A bot that talks to your customer. Voice agents, chatbots, virtual receptionists. | Automation |
| Conversation intelligence | Analysis of conversations that already happened between two people. | This product |
Published market research under the conversational AI heading is chatbot and virtual-agent research, sized around automated customer interaction (Research and Markets, 2026). Name your product conversational anything and four things follow:
- Your reps sell it as automation, because that is what the word means to them.
- Your customers expect something to answer their phone. It does not.
- Your first churn conversation is about a promise nobody made.
- Your legal exposure changes shape. The active litigation concerns AI that participates in the call. Post-call analysis sits in a smaller risk category, and there is no reason to name yourself into the larger one. See Chapter 08.
Avoid: conversational AI, conversational intelligence, AI receptionist, AI agent, voice AI, virtual agent.
Other language worth keeping out
| Avoid | Because | Use instead |
|---|---|---|
| Real time, live, during the call | Nothing happens during the conversation | Within minutes of hang-up |
| Listens, monitors, watches | Describes surveillance, and employees will read it that way | Surfaces, captures |
| Coaching, scoring, quality management | Not in the product. Different category and different compliance posture. | Leave it out |
| Routes, handles, resolves | It notifies a person. The person acts. | Alerts, surfaces to |
| Note taker | Undersells it. The tasks are the product. | Follow-through, commitments captured |
| Records your calls | Your platform already records | Works from your existing recordings |
Naming & Branding tab. Product name, one-line description, URL, DNS owner, email routing, sender name and address, SMTP details, logo files, hex colors, and whether you sell through your own channel partners.
Packaging and price
You buy at wholesale and set your own retail. We publish no retail price, for you or for anyone. Two decisions sit here: how it is packaged, and what it costs.
Packaging or add-on
| Packaged | Add-on | |
|---|---|---|
| What it means | Included in a seat you already sell, or in a new seat you create for it | A separate line a rep attaches to any seat |
| Sold by | Nobody. It ships with the seat. | The rep, on every deal |
| Why you would | You are losing accounts to providers with an AI story and want the answer in every seat you sell | You want to raise revenue per customer across a base you already hold |
| How reps position it | A new capability in the seat, and the reason it is there | Two or three questions. Are your people on the phone all day, do things get missed after calls, would you want to know when a customer is unhappy. |
| What we build for you | Seat positioning and competitive sheets | Talk track, objection cards, quote language |
Both work, and plenty of providers do both. Tell us which you are leading with before the meeting, because the enablement differs.
Expansion revenue is available either way. Expanded access is bought at the account level when a customer asks for it, and packaging the base seat does not close that door.
Setting your price
Across the partners we have priced with, retail has landed between $10 and $20 a seat. That band holds across providers of very different sizes, in different regions, with no contact between them. It is our observation rather than published research.
Your price is yours to set and we do not advise on it. The band above is context, not a recommendation.
Your wholesale rate improves as your book grows
Markup is a ladder rather than a flat rate, and the rung is set by your committed monthly volume across your whole book, not by any one customer's invoice. Your rungs and your entry terms are on your order form. Worth knowing before you set retail, because the next rung changes what you can afford to discount.
Working out the minutes
The minute mechanic is simple, and it is worth walking through once so it never becomes a surprise on an invoice.
Every seat carries 300 minutes
Pooled at the account level, not per user.
The pool is seats times 300
A 40 seat account has 12,000 minutes to spend across everyone.
Anything above the pool is usage
Billed per minute in arrears, rounded up to the next full minute.
One rate, always
Your overage rate is contracted on your order form and does not move as your markup ladder improves.
What to do with that before you price
Pull three months of minutes per account from your platform and divide by seats. That single number tells you which of three situations you are in.
| Average minutes per seat | What it means | What to do |
|---|---|---|
| Under 250 | The pool absorbs everything. Overage is effectively zero. | Price the seat and forget the meter. |
| 250 to 400 | Occasional overage, small amounts, month to month variation. | Either pass it through at your own rate, or pad your seat price a little and absorb it. |
| Over 400 | Consistent overage. Contact centers, dispatch desks, inside sales floors. | Sell a minute pack of your own alongside the seat, or set a higher seat price for that segment. |
The pooling matters here. A single heavy user inside an otherwise quiet account is usually absorbed. It is the whole account running hot that creates a bill.
Exhausting the pool does not stop processing, does not suppress alerts, and does not degrade anything. Minutes above the pool are simply metered. It is the first question most billing teams ask, and the answer is a good one.
Selling a minute pack of your own
There is no wholesale minute pack to buy. Your cost for anything above a customer's pool is the flat per-minute rate on your order form, and it does not change as your markup ladder improves.
What you can do is build a pack as a retail item of your own. If you sell on a promise that the customer's bill does not change month to month, per-minute overage breaks that promise. Price a block of minutes, absorb the variance against your contracted rate, and keep the breakage. It is the cleanest answer for a heavy account, and the arithmetic is entirely yours because your input cost is fixed.
Packaging & SKUs tab. Packaging decision, which seats it applies to, retail price, discount floor and approval path, commission treatment, average minutes per user, and whether you are selling a bucket.
Sales operations
A rep who cannot build the quote stops selling. Everything below is ready to copy into your quoting tool.
Items to create
Two items get you selling. The three expansion items are worth creating at the same time, because customers ask for them early and a request that has nowhere to go tends to die. Item 3 is optional and is a retail construct of your own rather than something you buy from us.
Item names are placeholders. Replace them with your own product name and adjust the wording to match how your catalog reads. Your wholesale rates are on your order form. Item 3 is the exception: there is no wholesale pack behind it. You price it yourself against your contracted per-minute rate.
Considerations for your sales process
| Item | Why it comes up early |
|---|---|
| Commission treatment | Attach rate follows the comp plan. Whatever you decide, reps need to know it before the first quote goes out. |
| Discount floor | Easier to set once, up front, than to unwind after a few quotes have set a precedent. |
| Approval path | If adding this line triggers an approval, whoever grants it should know the line exists. |
| Proposal template | Where the line sits on the quote affects how often it survives a customer's edit. |
There is an admin console. Your team creates customers and users in it directly, so activation does not route through us. How that works is covered in training rather than here, and the exact handoff for your first few accounts is agreed at the coordination meeting.
Packaging & SKUs tab. A status line for each of the six items, plus your billing platform and the address the monthly usage file goes to.
Billing
Two documents arrive each month: a usage detail showing what every one of your customers consumed, and an invoice. Both are shown below so your billing team can map them before the first one arrives.
| What we track | Licensed seats and total minutes, per customer. |
| What you receive | A usage detail and an invoice, on the first of each month. |
| How | Emailed to an address you name. |
| What does not exist yet | A billing platform integration or a billing API. If you will want a scheduled feed or an API pull, say so and it gets weighed against other partner demand. |
The usage detail
One row per customer, per item. This is what you rebill from. Each customer carries its own pool of minutes, so overage is rated per customer rather than across your book.
Issued 8/1/2026
| Customer | Description | Seats | Used | Pool | Rate | Total |
|---|---|---|---|---|---|---|
| Riverbend Dental Group | Seats | 42 | 14,980 | 12,600 | 7.00 | 294.00 |
| Overage minutes | 2,380 | 0.015 | 35.70 | |||
| Cole & Harper LLP | Seats | 26 | 6,540 | 7,800 | 7.00 | 182.00 |
| Stanton Auto Group | Seats | 88 | 31,420 | 26,400 | 7.00 | 616.00 |
| Overage minutes | 5,020 | 0.015 | 75.30 | |||
| Meridian Home Services | Seats | 84 | 22,940 | 25,200 | 7.00 | 588.00 |
| Total seat fees | 240 | 75,880 | 1,680.00 | |||
| Total minute fees | 7,400 | 111.00 |
Each customer is rated on its own pool. Unused minutes on one customer do not offset overage on another, and nothing carries forward month to month.
Minutes above the pool are billed at the usage rate and rounded up to the next full minute.
Illustrative figures at the entry rate. Rates are confirmed on your order form and improve at higher commitment levels.
The invoice
The invoice is the greater of your minimum monthly commitment or your actual seat billing, plus usage. In the month shown here, seat billing has passed the minimum, so the minimum no longer appears as a charge.
Issued 8/1/2026 · Due 8/31/2026
| Line | Description | Amount |
|---|---|---|
| Seat licenses | 240 seats at the rate on your order form | 1,680.00 |
| Processing minutes | Minutes above each customer's own monthly pool | 111.00 |
| Monthly minimum commitment | Not charged. Seat billing exceeds the minimum this month. | 0.00 |
Nothing recurring bills for your first two months. Seats begin selling in month two, and recurring billing including the minimum starts in month three. You never pay a commitment before you have had a real chance to sell against it.
Packaging & SKUs tab. Your billing platform, the destination address for the monthly file, and confirmation your billing team has reviewed the samples above.
Legal and contracts
This chapter and the next describe current law and active litigation. We are not your counsel, and neither is legal advice. What we can offer is where every provider before you has been surprised, and what your counsel will want in front of them.
The agreement you signed with us governs the relationship between us and you. The relationship between you and your customer is separate, and that is the one that matters, because your customer is the one talking to consumers. This is the item most likely to hold up a launch late.
Your customer agreement
Your existing terms may not permit what this requires. Your counsel is checking for three permissions:
- Can you store recorded conversations on behalf of your customer?
- Can you process those recordings for transcription and analysis?
- Can you use AI in that processing?
If any answer is no or unclear, an addendum is needed before the first seat is sold. We supply model language your counsel adapts into your own template.
The responsibility boundary
Compliance with recording and AI law in the jurisdictions where your customers operate is your customer's obligation and yours. It is not ours. This is in your agreement with us and it needs to flow through to your agreement with them. Chapter 08 covers what your customers do about it.
Insurance and liability
Adding conversation processing to your service changes your risk profile, and it is worth confirming your existing cover still fits before the first customer goes live rather than after.
Three things for your team or your broker to look at:
- Technology errors and omissions, and cyber liability. Confirm the policy responds to claims arising from processing customer conversation data, and that the limits are appropriate to the size of the base you intend to attach.
- Your customer agreements. Confirm the limitation of liability and indemnity provisions still hold once conversation processing is part of what you deliver.
- Vendor flow-through. Confirm that a claim brought against you by a customer does not, by the terms of your own paper, pass straight through to your suppliers, ourselves included.
We are not prescribing limits or naming a carrier. This is a checkpoint on the launch plan so it is not discovered late.
Healthcare
We have not completed a HIPAA audit, and we would rather say so now than have you find out later. We serve healthcare accounts, and we would rather they were not the first thing you lead with while that is true.
If you intend to sell into healthcare at all:
- We need a subcontractor business associate agreement with you. One agreement covers your whole base, not one per customer.
- Your contract with us already requires that protected health information is not sent to us before that agreement is signed.
- The chain has to be unbroken: your healthcare customer to you, and you to us. If your instance is hosted by Crexendo, confirm the agreement with them as well.
The agreement is staged after the reseller agreement deliberately, so signing us does not mean reviewing two documents at once.
What we do with the conversation
The next chapter tells you to ask every vendor in this category, including us, what they do with your customer's conversations. Here is our answer, so you do not have to ask.
Card data is redacted before anything is analyzed. Personal and health information is stored and analyzed rather than removed, because content removed from a transcript is content no analysis can read. It is held under scoped access, with retention split on identifiability: identifiable content follows the customer's own requirements, and derived, unidentifiable content is governed by us. We retrieve a copy of the recording, hold it under the stated retention schedule, and serve playback from it. We do not train models on customer conversations. The sub-processors that see conversation content are available on written request.
Disclosed vendors
If you maintain a vendor list under a customer agreement, a certification, or a procurement obligation, we belong on it. Disclosure is also the most effective defense against the litigation described in the next chapter.
Legal & Compliance tab. The three permissions, addendum ownership, certification status, data residency, existing data processing agreements, insurance, healthcare intent, PCI handling, and your current disclosure language.
Recording, consent, and disclosure
Your reps will be asked about this in the field. This chapter gives them something better than a shrug. It is current as of August 2026 and the ground is moving, so check the review date at the top of the page.
Two things that surprise almost everybody
| Expectation | Reality |
|---|---|
| The new state AI laws are the risk | They mostly do not reach this product. California's SB 942 concerns watermarking of synthetic media by large generative providers and became operative August 2, 2026. AB 2013 requires generative developers to publish training data summaries. Colorado's AI Act, which addresses consequential decisions in areas like employment and credit, was scaled back and pushed to January 1, 2027. None of that describes summarizing a completed sales call. |
| Wiretap law is settled and old | It is old, and it is where all of the litigation is. |
Sources: Cooley, April 2026; Hunton, May 2026.
One-party and all-party consent
Federal law is the floor and requires one party to consent (The Lyon Firm, June 2026). States may be stricter and a number are.
The published lists disagree with each other, which is itself the finding. Counts range from nine states to fifteen depending on how the compiler treats statutes that split between telephone and in-person, or that have been narrowed by case law.
| Category | States |
|---|---|
| Consistently listed as all-party | California, Delaware, Florida, Illinois, Maryland, Massachusetts, Montana, New Hampshire, Pennsylvania, Washington |
| Split, narrowed, or disputed | Connecticut (all-party for telephone under the civil statute), Nevada (all-party for telephone), Oregon (all-party in person, one-party by telephone), Michigan (statute reads all-party, courts recognize a participant exception), Vermont (no statute) |
| Everything else, plus DC | One-party, matching the federal floor |
Three compilations that do not agree: Recording Law, SIPNEX, Layer3 Labs, August 2026.
When participants sit in different states, the strictest applicable law tends to control. Any customer with callers in more than one state, which is nearly all of them, arrives at the same policy anyway.
Disclose on every conversation, in every state, for every customer. The list explains why the policy exists. It is not the policy.
The risk that is genuinely new: the vendor as a third party
Plaintiffs are arguing that an AI vendor processing a call is not a tool of the business but a separate third party listening in. Claims run under the federal Wiretap Act and state statutes such as California's Invasion of Privacy Act (Holland & Knight, May 2026).
| When | What happened |
|---|---|
| 2026, California | A federal court allowed claims against a call analytics vendor to proceed, holding that a vendor's capability to use conversation data for its own purposes was enough to plead eavesdropping. Two other California district decisions aligned. |
| April 2026 | Plaintiffs survived dismissal where an AI transcription and analytics tool was used on service calls without customers being told (Paul Hastings). |
| January 13, 2026 | An Illinois federal court dismissed a wiretapping claim against a communications platform provider and its customer. |
| July 30, 2026 | A California class action targeted an AI voice ordering vendor over calls to a restaurant chain. The vendor is the defendant and the restaurant is not, because the complaint frames the vendor as acting independently for its own benefit (Mondaq, August 2026). |
Three conclusions for your team:
- Nondisclosure is the common thread. Case after case turns on nobody being told a third party was processing the conversation. Disclosure is the defense, not a nicety.
- Whether the vendor uses the data for its own purposes is doing real work in these rulings. Ask every vendor in this category, including us, whether they retain your customer's audio and whether they train models on it. Get the answer in the contract.
- Relief is not arriving from the legislature. California's SB 690 was expected to create a broad commercial purpose exemption. The July 2026 amendment dropped it and narrowed the bill to pen register claims, leaving the core wiretap provisions untouched (National Law Review, July 2026).
Biometrics
Voiceprints are regulated biometric identifiers in Illinois, Texas, and Washington. Illinois is the loud one because its statute carries a private right of action (NJ SaaS Law, July 2026). Texas has no private right of action, and its attorney general has still recovered more than $2.7 billion combined from two technology companies (RatedWithAI, July 2026).
The question is narrower than it sounds: does the system create a voice template. Some courts have found that telling speakers apart in a transcript can create a voice embedding sufficient to trigger the Illinois statute. In May 2026 nine coordinated class actions were filed in Illinois against major technology companies over voiceprints allegedly extracted from recordings to train voice models (American Bar Association, July 2026).
Does the pipeline create or persist a speaker embedding, and for how long. A vendor who cannot answer in one sentence has answered.
Where there is no IVR
The most common question your reps will get, and the one that stalls deals: a call center announces recording through an IVR, but most businesses do not run their people through one. Someone dials a direct number, a person picks up, and nothing plays.
The law requires consent. It does not require an announcement. An announcement is one way to obtain consent and it happens to be the way call centers do it. On a direct-dial line consent is obtained another way, and most businesses use two or three of the options below together.
| Mechanism | What it looks like |
|---|---|
| Announcement on the line | A short recorded notice when the call connects on the seated user's number, rather than on a call center path. |
| Scripted verbal disclosure | One sentence at the start of the call. The most common answer, and it costs nothing to implement. |
| Standing notice to known contacts | A clause in the service agreement, a line in the email signature, a line on quotes and work orders, and a one-time notice to the existing contact list. Consent for the relationship rather than for each call. |
| Choose who is seated | Not everyone needs a seat. A business can enable inside sales and service coordination and leave field staff off entirely. |
| Record on demand | Recording switched on by the user once consent is given. A conversation that is not recorded produces no record at all. |
One more point worth making to a customer who raises this: the obligation exists the moment they turn on call recording. Adding conversation intelligence does not create it. What it adds is a second thing to disclose, which is the processing.
There is a two-page compliance brief covering this in full, with model language for the call, the contract, the email signature, and the employee notice. It is written to be forwarded to an end customer, and we can put it in your brand.
Model disclosure language
Starting points, not finished language. Your counsel approves them before they go live.
This call may be recorded and processed using automated tools to create summaries and follow-up notes. If you prefer not to be recorded, please let us know and we will assist you another way.
Before we start, I want to let you know this call is recorded and we use an automated tool to write up notes afterward. Is that all right with you?
Two design points matter more than the wording:
- Disclose the processing, not only the recording. The complaints above are about undisclosed processing by a third party. "This call may be recorded" was written for a world where recording was the whole activity.
- Your customer decides what happens when a caller declines. A disclosure with no alternative is not consent, so their call path needs a route that does not record. Worth stating plainly to them: a conversation that is not recorded produces no record in the platform, so nothing about it will appear in summaries, alerts, or reporting.
What to tell your customer about their own staff
Two facts answer most of what employees raise, and both belong in front of your customer before go-live:
- Alerts go only to the people the business nominates. There are two triggers in the base seat, and neither is a performance measure.
- Individual conversations are not scored, graded, or ranked.
An employee notice template is in Chapter 15.
IT and platform
We retrieve the recording and the call detail record from your platform as soon as a conversation completes. This chapter covers what makes that work on the first attempt.
API keys
One API key per enabled customer. It limits exposure if a key ever has to be recycled, and it means a single customer issue never touches the rest of your base. If your operating model genuinely cannot support it, raise it at the coordination meeting and we will work out an alternative.
Keys come to security@tresic.cloud by secure transfer: a password manager share, a one-time secret link, or encrypted email. Never in a chat message, a ticket, or a spreadsheet, including the launch workbook. Tell us before you rotate a key rather than after.
Platform connection
For NetSapiens-based platforms we confirm the hosting model, API version, allow listing, and the recording configuration once, and it holds for every customer after that. Older API versions we have not worked with before are the one thing worth flagging early rather than at activation.
NetSapiens is the connector that is live today. Other platforms are connector work we scope with you, and each connection is handled individually: some combination of call detail retrieval, recording retrieval, and an API connection. Raise it early rather than at activation so it can be scoped against your launch date.
Recording
Recording must be enabled for every user or line carrying a seat. Beyond that we need to know the retention period, whether the recording API is publicly reachable, and any restrictions in place such as encryption or signed URLs.
Other platforms in your base
Tell us what else your customers run on: SIP-only endpoints, Microsoft Teams, SIPPIO, Intermedia, or anything else, along with whether those calls route through your primary platform today. Native Microsoft Teams call support is not available at launch, so if a meaningful part of your base is on Teams it is worth saying so and having it weighed against other demand.
Identity and user management
Single sign-on is available in early Q4 2026.
API access is included in the seat price. What it covers is an event stream out and a conversation record read: webhook push when a conversation is processed and when an alert fires, plus query and retrieval of the conversation record. Every read is scoped to the user the integration authenticates as. Your team builds against it with a user credential. Customer-managed API keys are not available yet, and user and customer provisioning is done in the admin console rather than through the API.
Setup tasks
| Task | Owner |
|---|---|
| Add our egress addresses to your allow list, if you use one | You |
| Add your addresses to our allow list | Us |
| Add the DNS record for your chosen URL | You |
| Enable your URL and configure branding | Us |
| Provide SMTP details, if mail leaves from your domain | You |
| Configure mail routing and sender name | Us |
| Send your platform API key | You |
| Configure the connection and confirm the first conversation processed | Us |
Platform Config tab. Connection details, API version, allow listing, key issuance, recording configuration and retention, capture scope, other platforms in your base, and identity intent.
Solutions engineering
Your solutions engineers write the proposals and run the implementations, which makes them the internal owners of this product in practice. Their track goes deeper than the sales track and covers configuration, not just positioning.
The configuration decisions on every account
Three answers before provisioning. They belong in the proposal rather than being discovered at go-live.
| Decision | Detail |
|---|---|
| Who gets a seat | A seat ties to a specific extension. Everyone with a seat gets After Call for their own conversations. |
| Who receives alerts | Alerts are set at the organization level and delivered to the people the business nominates. The two in the base seat appear in the console as Upset Customer and Customer Threat. Recipients are usually the people who can act on the issue, rather than the person who had the conversation. |
| Who sees what | Some businesses want full transparency and some do not. Settle it with the customer before provisioning, because changing it later is a conversation with their staff. |
A twenty-two person healthcare front office gives every phone agent their own summaries, with alerts going only to managers. A twenty person auto service business gives nobody but the owner access at all. Both are correct, and the configuration conversation is short once you know to have it.
There is a third named alert, Customer Dispute, that the base seat does not include. When it matches, the customer sees that something fired without seeing what. That is the moment to raise expanded alerts, and it arrives on its own rather than needing a rep to prospect for it.
What your SEs need available
- A demo account they can show without calling us.
- Their own seat, running on their own conversations. Nothing demonstrates this as well.
- The admin console, so they can configure a customer end to end.
- API and webhook documentation, for the customers who ask.
Training tracks
| Track | Who | Covers | Length |
|---|---|---|---|
| Sales | Reps and sales managers | What it is in one sentence, objections, how to quote it, competitive positioning | 60 min |
| Solutions engineering | SEs and architects | The sales track plus configuration, deployment patterns, what the customer sees, the APIs, and what breaks | Half day |
| Admin console | Provisioning and Tier 1 support | Creating customers and users, alert recipients, the configuration surface, common errors | 90 min |
Configuration training is scoped to what you are actually selling.
User List tab. Everyone internal who gets a seat, with extension, domain, email, department, and which training track they belong in.
Product and feedback
Expansion is sellable from day one. Your reps can quote every expansion item alongside the base seat, and your solutions engineers field the detail when a customer wants to go deeper.
Nothing here waits on a roadmap. The three expansion items in Chapter 05 are live products with published wholesale rates, and they sit in your catalog next to the base seat. What tends to happen is that customers reach for them from inside the product rather than being sold them up front, which is why reps stay on the attach and the expansion conversation arrives on its own.
Where product feedback goes
| What | Where |
|---|---|
| Feature requests, gaps, anything a customer asked for and could not get | support@tresic.cloud |
| Bugs and anything not behaving as documented | support@tresic.cloud |
| Roadmap conversations, architecture questions, anything strategic | Direct with our CTO. You will have that contact from the coordination meeting onward. |
Everything sent to support is logged and read. Nothing gets lost because it arrived through the wrong door.
What is most useful to us
- What customers ask for that they cannot get today.
- What your reps get asked in the field that they cannot answer.
- Which of your customers would talk to us about their experience, once they have one.
- Where the product gets in the way of your own operation.
Support
You own the customer relationship, so you own the first call. We back you.
| Layer | Who | Covers |
|---|---|---|
| Tier 1 | You | Login and access, who sees what, adding and removing users, alert recipients, how to use the product, billing questions |
| Tier 2 | You, then us | Configuration that will not take, conversations not appearing, platform connection issues |
| Escalation | Us | Platform faults, data questions, anything touching security |
The three most common early tickets
- No conversations are appearing. Nearly always recording not enabled on that user or line.
- I got an alert and I do not know why. Alert recipients set without telling the recipient. The templates in Chapter 15 handle this.
- Why can my manager see this. A configuration conversation that did not happen at provisioning. See Chapter 10.
All three are preventable at setup, which is why the admin console training and the communication templates matter more than they look.
Contacts tab. Tier 1 owner and escalation contacts on your side.
Marketing
We build the assets in your brand. You publish and send them. Our name appears on none of it.
What we deliver, what you deliver
The assets below are produced as part of your onboarding.
| We produce, in your brand | You do |
|---|---|
| Base announcement emails | The send, from your system, to your list |
| An HTML block for your website | Drop it in and publish |
| Product marketing one-pager slicks | Put them in front of your reps and your customers |
| Portal banner artwork | Place it |
| Competitive sheets against the named platform vendors For internal rep use, not for customer distribution | Get them in front of your reps |
| Sales talk track and objection cards | Run the sales meeting |
The sequence that works
Your own team first
Every rep and SE runs it on their own conversations. Not a demo, their real calls.
SEs go deeper
They write the proposals and run the implementations.
Announce to your base
Your existing customers are the whole opportunity.
Then net-new
By now your reps have references, screenshots, and their own experience.
The first step is the highest-return thing you can do and it costs only the seats. A rep who has watched their own conversation come back as finished work sells it differently from one reading a card.
Why now, for the net-new conversation
Analysts describe platform intelligence rather than feature count as the axis this category now competes on (Future Market Insights, February 2026), and at least one major UCaaS vendor shipped voice summaries and conversation intelligence into its platform this year (Custom Market Insights, July 2026). Your customers are being told a story. This is you telling one back.
An HTML block and a portal banner is the whole website ask. Nobody is asking you to rebuild anything. The block exists mostly so a rep has somewhere to point.
Contacts and Launch Plan tabs. Who can publish an HTML block to your website, who sends to the base, who owns the portal, plus your target announcement date.
Message library
Eight lines you can use anywhere, in any combination. These are the same lines running through the assets we build for you, so anything you write yourself stays consistent with what we hand over.
Lead lines
The headline for an announcement to customers who already have the phone. Carries the whole idea without explaining anything.
For anyone whose first question is what they have to change. The answer is nothing.
The claim no platform competitor can make, because every one of them requires the customer to move.
For an owner who wants the business case in one line before the demo.
Capability lines
The summary is table stakes. The tasks are the product, and this line puts the emphasis in the right place.
For the moment a customer asks what happens when something goes wrong on a call.
For the owner or manager who has never had a way to know what is being said all day.
Useful when explaining that alerts go to the person who can act, not to everyone.
No line here claims anything happens during the call, and none of them describes coaching, scoring, or monitoring. Both are outside the product and both create problems with your customers' employees. Chapter 03 has the full list.
Email templates
Five emails cover the whole launch. Two go to your own team, three go to customers. Fill in the bracketed fields, cut anything that does not apply, and send.
Team,
Starting [date] we are adding [product name] to our portfolio. It attaches to the seats we already sell and gives our customers a summary of every conversation, the follow-ups captured as tasks, and alerts when something needs attention.
Two things to know now.
You are getting it first. Everyone on the phone here will have it running on their own conversations before we sell a single seat. You will receive a login from [sender address] in the next [timeframe]. Set your password, then use the phone the way you already do. Your summaries will start arriving the same day.
Nothing changes about how you work. No new app to open, no new process. It runs on our existing phone system.
What we can see, plainly: [alerts for the two defined triggers go to X and Y only]. Individual conversations are not scored, graded, or ranked.
[Name] is running the rollout. Questions to [channel or address].
From [date], [product name] is on the price book.
What it is, in one sentence. [Your one-liner]
How it is sold. [Included in the [seat] / An add-on at $X per seat per month]. Two items on the quote: the seat and processing minutes. Minutes are billed only above the included allowance.
Commission. [Treatment]
Discount floor. [Floor and approval path]
Before you quote it, confirm the account has call recording enabled. That is the one hard requirement.
Your own seat is being set up now. Use it for two weeks before you pitch it.
Training is [date and time]. Materials are at [location].
Hi [first name],
We have added something to [your service] that will change how your team works: [product name].
After every conversation, your team gets a written summary and every commitment made on the call captured as a task, ready to assign. Nothing gets lost between the call and the follow-up. Managers get notified when a conversation needs attention, and you can see what your whole business is hearing without listening to a single call.
It runs on the phones you already have. Nothing to install, nothing to migrate, and nothing new for your team to learn.
[Learn more / See what it costs / Talk to your account manager] at [link].
Hi [first name],
[Product name] is now active on your account. Here is what happens next.
Your team. [Number] people have been set up. They will receive a login from [sender] and can sign in at [URL].
Alerts. These currently go to [names]. Tell us any time you want that list changed.
Before your team starts, please send them the note below so nobody is surprised by their first summary. We have written it for you.
Summaries start today. Conversations from today onward will appear the same business day.
Questions go to [support channel].
Starting [date], conversations on company lines are processed to produce a written summary and a list of follow-ups. You will get yours by email after each call, and you can find past conversations by searching.
What this is for. Follow-through. Nothing gets lost between the call and the task.
What managers see. [Managers receive an alert only when a conversation triggers one of two defined signals: a frustrated customer, or a customer threatening to leave or escalate. Alerts go to [names].]
What is not happening. Individual conversations are not scored, graded, or ranked. There is no coaching feature and no performance measure.
Callers are told. Our recorded-call disclosure now covers automated processing as well as recording.
Questions to [name].
The first four weeks
Seven gates. Each has evidence attached rather than an opinion. The full task list, with owners and dates, is the first tab of the workbook.
| Week | Gate | Evidence it is closed |
|---|---|---|
| Meeting | Contractable | Order form executed, model addendum language with your counsel |
| 1 | Provisionable | Platform connected, branding live, first internal user processing conversations |
| 2 | Orderable | Items in the quoting tool, test quote produced |
| 2 | Billable | Billing items created, samples reviewed by your billing team |
| 2 | Demoable | Every rep and SE running it on their own conversations |
| 3 | Compensable | Commission treatment decided and communicated |
| 3 | Contractable | Addendum executed into your customer template |
| 4 | Marketable | Website block live, portal banner up, base announcement sent |
| 4 | Launched | First quote carrying the line |
Launch Configuration Workbook
Every question in this guide, in one file, split by team.
Make a copy, complete it, and share it back with your Tresic launch contact five business days ahead.
Sources
Every external claim in this guide, with its date. All sources are dated within six months of August 2026. Chapters 07 and 08 are re-checked quarterly at minimum.
Wiretap, eavesdropping, and AI vendor liability
| Source | Date | Used for |
|---|---|---|
| Holland & Knight | May 2026 | The capability standard in the call analytics case; the January 13, 2026 Illinois dismissal; AI vendors framed as third-party eavesdroppers |
| Paul Hastings | April 2026 | Plaintiffs surviving dismissal where AI transcription was used without disclosure |
| Mondaq | August 2026 | The July 30, 2026 filing naming the AI voice vendor rather than the brand |
| The Lyon Firm | June 2026 | The federal one-party floor; the plaintiff-side view of active litigation |
| National Law Review | July 2026 | The July 2026 amendment dropping the commercial purpose exemption |
| Covington | July 2026 | Narrowing to pen register claims; the January 1, 2027 operative date if enacted |
Consent by state
These compilations contradict each other, which is why the guide recommends universal disclosure rather than a state-by-state policy. No single one should be treated as authoritative on any individual state.
| Source | Date | Position |
|---|---|---|
| Recording Law | 2026 | Twelve all-party states; Michigan classified as one-party under the participant exception |
| SIPNEX | 2026 | Eleven clear all-party states including Nevada; four more mixed; treat fifteen as all-party in practice |
| Layer3 Labs | August 2026 | Statutory citations for the split states |
| Nimitai | May 2026 | Twelve all-party states with statutory citations |
Biometric privacy
| Source | Date | Used for |
|---|---|---|
| American Bar Association | July 2026 | The nine coordinated Illinois class actions filed May 2026 over voiceprints used to train voice models |
| Andrew S. Bosin LLC | July 2026 | Three states with standalone biometric statutes; speaker diarization potentially creating a regulated voice embedding |
| RatedWithAI | July 2026 | Texas has no private right of action; the attorney general recovered more than $2.7 billion combined from two companies |
State AI statutes
| Source | Date | Used for |
|---|---|---|
| Cooley | April 2026 | AB 2013 effective January 1, 2026; SB 942 delayed to August 2, 2026 by AB 853 |
| Hunton Andrews Kurth | May 2026 | SB 189 signed May 14, 2026, delaying the Colorado AI Act to January 1, 2027 and scaling it back |
Market context
| Source | Date | Used for |
|---|---|---|
| Future Market Insights | February 2026 | Platform intelligence rather than feature count as the competitive axis |
| Custom Market Insights | July 2026 | A major UCaaS vendor shipping voice summaries and conversation intelligence into its platform |
| Research and Markets | 2026 | Evidence that published conversational AI research is chatbot and virtual-agent research |