AI agents installed in your business, with a person in charge
We pick one repetitive workflow, build an agent for it, connect it to the tools you already use, and agree in writing when it hands the work to a person.
Who this is for
Owner-led service firms (law, accounting, agencies) and operations teams of small businesses in the US. It usually fits when:
- your team answers the same questions by hand every day;
- new inquiries wait hours for a first reply, and follow-up depends on someone remembering;
- documents arrive by email in different formats and someone retypes them;
- the owner is the bottleneck for routine decisions.
It is not a fit yet if the workflow changes every week, has no clear owner, or a mistake cannot be caught before it reaches a customer. We will say so in the audit and suggest fixing the process first.
Two example workflows
These are example workflows, not client results.
Example A. Inbound inquiries and follow-up for a service firm
- The agent reads new inquiries from your web form and inbox, asks for missing details, and sorts each one: new client, existing client, not a fit.
- It drafts the reply and the follow-up, and offers the person a time slot in your calendar.
- A person approves: every reply during the first weeks; anything about price, contract terms or legal questions; anything the agent marks as unsure.
Example B. Document intake and routing
- The agent receives documents by email or upload form, checks what is missing, and asks the sender for it.
- It extracts the key fields into your system and routes the case to the right person.
- A person approves: anything with low confidence, any exception, anything that changes money or a client record.
How an engagement works
- Free AI audit, no call needed. Accept a mutual NDA at audit.kis.software, then email us one data export (calls, leads, orders or CRM) and three numbers. We send back a one-page report: where your business is losing money each month, 3 specific fixes, and an honest view of whether an AI agent fits the first fix. No obligation.
- Written design. What the agent does, what it never does, handoff and escalation rules, and how we measure success. You approve it before we build.
- Paid pilot: build and connect. The agent for that one workflow, connected to your email, forms, calendar, CRM or document storage, through official integrations where the tools offer them.
- Testing on your cases. Before the agent touches live work, we run it on a set of past examples and tricky cases, and review the results with you.
- Supervised launch. The agent starts by drafting, a person approves. We widen what it does on its own only where the results support it.
- Handover. Documentation, admin access, how to pause the agent, and a short training for your team. Support after launch is agreed in the proposal.
What stays with your team
- One owner on your side who knows the workflow and decides on exceptions.
- Access to the systems the agent works in, through accounts you control.
- Reviewing and approving the agent’s work during the supervised period.
- Final decisions on pricing, contracts, legal questions and money.
- Telling us when the process changes, so the rules change with it.
Handoff and escalation rules
We write these rules with you before the build. Typical starting rules:
The agent hands the work to a person when:
- the customer asks for a person;
- the request involves price, contract terms, a complaint or a legal question;
- it cannot find the answer in the sources you approved;
- it is unsure, below the threshold we agreed;
- the same customer writes again after getting an answer.
How the handoff works: the person gets the full conversation, the agent’s draft and the reason it stopped, in the channel your team already uses (email, Slack, Teams or a CRM task). The customer is told that a person will reply, and when.
The agent never: promises discounts, dates or outcomes; makes payments; deletes records; sends anything outside the approved templates during the supervised period.
Data security, in plain words
- The agent gets only the access its workflow needs, read-only where possible.
- We work in accounts you own, and you can revoke our access at any time.
- Before the build, we tell you which AI model providers will process your data and on what terms. Where the provider offers it, we use settings that keep your data out of model training.
- Every action the agent takes is logged, so you can see what it did and why.
- You can pause the agent at any time.
- We sign an NDA on request.
How it works
- New inquiry or document
- Agent sorts and drafts
- Rules check
- Person approves or takes over
- Reply sent, record updated
Demos on this page will use synthetic data only and will be labeled as such. We never show client data.
How the free AI audit works
No call needed. Three steps:
- Accept a mutual NDA at audit.kis.software, before you share anything.
- Email us one data export (calls, leads, orders or CRM) and three numbers: your average order value or margin, the share of inquiries that turn into sales, and the share of new customers.
- Get a one-page report by email: where your business is losing money each month, 3 specific fixes, and our honest view of whether an AI agent fits the first fix, what it would take and what could go wrong.
If you want to go further, we send a written pilot proposal with scope, price and timeline, and we talk then.
FAQ
How long does it take?
The audit needs no call: you send one export and get a one-page report by email. For a pilot, it depends on the workflow and the systems involved: the timeline is in the written pilot proposal.
How much does it cost?
The audit is free. The price of a pilot depends on the workflow, the integrations and the support you need; you get it in the written pilot proposal, before any commitment.
Will the agent replace my staff?
We design agents to take repetitive steps off your team. People keep the decisions.
Which AI models do you use?
We choose the model provider for each workflow and tell you which one and why before the build.
Can we start small?
Yes. One workflow is the start we recommend.
Who we are
KISS SOFTWARE, LLC is a software company based in Laguna Beach, California: a small core team plus vetted contractors we have worked with before. Before we focused on AI, we built web and mobile apps, CRM systems, Telegram bots and security testing for clients in the US and Europe. Rated 5.0 on Clutch (4 reviews).
Have a workflow your team repeats every day?
Send us one data export and get a free one-page AI audit. No call needed.