# A proactive agent working for your business in about two months. Alknoma, in the proactive agent space, builds an AI agent that takes one job off your team's hands. It starts the work itself and sees it through, without anyone having to prompt it or look after it. The first step is a free 30-minute call. ## AI is paying off for tech companies. For most other businesses, not yet. Apollo's chief economist found AI has "delivered zero financial payoff to corporate America outside of tech" ([Yahoo Finance, October 2026](https://finance.yahoo.com/markets/stocks/articles/apollo-torsten-slok-warns-ai-125709125.html)), and small firms use it least: 17% of US firms with fewer than 20 staff, against 30% of large ones ([Federal Reserve Bank of Minneapolis](https://www.minneapolisfed.org/article/2026/ai-adoption-in-business-grows-steadily-but-unevenly)). The promise was that anyone could build anything. For many service businesses it has meant more competition, more noise, and owners pushed into building software they do not know how to look after. We think that is one of the reasons the payoff has not reached them yet. ## Four usual routes, and what can go wrong with each Each can work. Each carries its own risk. ### Hire an AI team An AI engineer, a senior software engineer and a product manager, at the pay it takes to hire people who can do it. Good for: - Built exactly around your business, and all of it is yours. What can go wrong: - Two to four months before anyone starts, then the build, with the payroll running throughout. - The knowledge sits with a few people. When one leaves, part of the agent leaves with them. - If the agent never pays off, the business still carries the salaries. Cost: UK Over £475,000 in the first year; US Over $720,000 in the first year. ### Build it yourself on an AI platform AI companies say anyone can build an agent on their platform. Someone on your team does the building. Good for: - Cheap to start, and built by the people who know the work best. What can go wrong: - The real cost comes later: usage that grows with every run, and your team's time to keep it going. - It acts wrongly and nobody notices: the wrong message to a customer, a reminder that never went out. - Company data goes into tools nobody checked. One in five organisations has had a breach through AI used outside IT's view (IBM, 2025). - About 1 in 3 AI tools built in-house ever reach use (MIT, 2025), and when the builder moves on, nobody can fix it. ### Use a tech company's own agent The big platforms offer ready-made assistants and agents inside the tools you already pay for. Good for: - Quick to switch on, with nothing to build. What can go wrong: - It works their way, not yours, and bends little to how your business actually runs. - You are tied into their ecosystem, and leaving gets harder every month. - Most still wait to be asked, so your people still follow up and check: about 4.5 hours a week fixing AI's work (Zapier survey). ### Hire a general AI studio Agencies that build whatever AI project comes through the door. Good for: - Someone else does the build, to an agreed scope. What can go wrong: - They rarely know your industry, so the agent misses how the work really runs. - Agent patterns change month to month, and a generalist is often a step behind. - They build it and leave. Keeping it running becomes your problem, and that is where the real cost sits. ## An agent your team can keep running. We build an agent for one job, with what we know of your industry and the newest ways agents are built. Then we make sure your team knows how to keep it running, and give you the free, source-available tools we use to check every change. We have been building agents like this since we introduced Ayven in January 2025. - **Lower cost.** No salaries to carry, and it only works when there is something to do. - **Better reliability.** Tested over weeks of practice runs before launch, and again after every change. - **Easier to maintain.** Your team learns how to look after it, and our tools check every change for them. - **Use Ayven** (Ready-made). Our own agent for Slack and Microsoft Teams. Give it the outcome: it plans the work, hands each part to the right person and sees it through to done. We shape it around your team. See Ayven: https://www.alknoma.com/ayven - **Have one built** (For one job of yours). We build an agent for one job in your business, test it, hand it over working and teach your team to keep it running. It can run in your own cloud, by contract. Book a free call: https://calendar.app.google/ZqjSZAbchWBTL6dw8 ## Start with a free call. Keep everything from it. Thirty minutes on Google Meet, with no commitment. Afterwards we send you: - **A written plan of the job.** Who an agent would ask, what it would wait for, and when the job is done. - **A shortlist for your business.** The jobs worth handing to an agent first, and the ones that are not. - **A guide to build it yourself.** The steps, and Minutehand, our free tool for testing it. ### What we would like to hear from you - How your business runs today, and the job that keeps slipping. - What you want AI to do for you in 2026 and 2027, and what you have tried so far. - Where an agent that acts on its own would help you, and where you would never want one. There is no contract and nothing to sign. We would simply love to know how businesses like yours want to use proactive agents. ## About two months from the call to a working agent. 1. **Built and practised.** We build the agent in the tools you already use, then run it through weeks of practice in a few seconds. An agent's mistakes often show up days in, not in its first reply, so that is where we look. 2. **Live with your team.** It goes live once you have seen it pass, about two months after the call. What you get: - **Your agent, at work.** Doing the job every day, in the tools you already use. - **Proof that it works.** Weeks of practice runs before launch, and what we fixed. - **A record you can check.** Everything it did, and why, whenever you want to look. - **A team that can keep it running.** We show whoever looks after it how, and leave them the tools to check every change. ## One pattern, two very different jobs One works inside a team, the other with customers. Neither waits to be asked: both decide for themselves when something is worth doing. - **Ayven.** Inside a team: given an outcome, it plans the steps, asks each person for their part, files what it cannot do as a ticket and watches it to Done, over weeks if it has to. ayven.ai: https://www.ayven.ai - **A prospecting agent.** With customers: a demo on a studio's website that watches what each visitor does and decides on its own when to say something. It says less when its messages go unread, checks every claim against its sources, and was tested in Minutehand before it went live. Read the code: https://github.com/HHMingD/prospect-agent ## Before you book ### What does it cost? It depends on the job, so we scope it on the free call. It costs less than hiring a team, and there are no salaries to carry. ### Do we need technical staff? No. We show whoever looks after it how to keep it running, and our tools check every change for them. ### How do we know it will not make things up? Before launch we check everything it said and did across weeks of practice runs, and flag any fact it was never given. It goes live once you have seen it pass. ## Tell us about the job. Pick a time for a free 30-minute Google Meet. There is no commitment, and the plan, the shortlist and the guide are yours whatever you decide. You get a calendar invite with the Meet link straight away. Howard Chen, our founder, runs every call. Book a Google Meet: https://calendar.app.google/ZqjSZAbchWBTL6dw8. Email: contact@alknoma.com.