HERENDILApplied AI Book a call

Herendil · Applied AI consulting and builds

A company brain for your people and your agents.

Herendil helps companies put AI to work on what they already know. We work out where agents and models fit your team, then build the agents, the inference behind them and the company brain they all draw on.

Docs01Pricing policy, v3
Tickets02#482 Refund stuckOpenP2
Chats03“Who owns the EU launch?”
Decisions04Ship EU before US
PeopleIdle
AgentsIdle

A company brain takes in a company’s docs, tickets, chats and decisions. People and agents ask it questions and get answers from what the company already knows.

Press Enter to load this card into the brain or take it out. Arrow keys move it.

Arrow keys move it.

01 · What we do

Advice and building, from one small team.

Most projects need both. Press a key to see each kind of work.

Agents with one job each

We build agents for a defined piece of work, such as triaging support tickets or preparing the month-end close. Each one works through your existing tools and permissions, and a person approves anything with consequences.

You get

A working agent, the test cases it has to pass, and a log of every action it takes.

Inference that fits the work

We choose the models, decide where they run and track what each answer costs. The result is fast enough for the task and cheap enough to run every day.

You get

A setup your engineers can read, with a cost per task you can check.

A company brain

We connect the places where your company keeps what it knows, so people and agents can search it and ask it questions. Access follows the permissions you already have, and the brain stays current as the work changes.

You get

One source that every agent and every teammate draws on.

Advice before code

We look at how your team works today and name the few places where AI would save real time. We also say where it would not, and what to try first.

You get

A short written plan you can act on, with or without us.

02 · How we work

Four stages. Stop after any of them.

Whatever we made up to that point is yours to keep.

Slide to a stage.

  1. 01

    Look

    We sit with the people doing the work and read the systems they use. You get a short list of jobs worth automating and the data each one needs.

  2. 02

    Try

    We build a small version on your real data and compare it with how the job gets done today. If it isn’t better, we say so and stop.

  3. 03

    Build

    We make it ready for daily use, with access control, evaluations, logging and cost limits. It runs in your own accounts.

  4. 04

    Hand over

    Your team gets the code, the documentation and a walkthrough. We stay on call for as long as you want.

03 · Notes

Notes from the bench

Short pieces on agents, inference and company brains. The first ones are being written now.

04 · Start here

Book a call

Thirty minutes, no slides. Bring one job you would like an agent to take on, or a question about your data. You’ll leave knowing whether it’s worth building and roughly what it would take.

Find a time The calendar link goes live on this key soon.