Build with AI from day one
For founders creating a new organization: define the work, knowledge base, agent responsibilities and human decisions before adding more tools.
Begin with the first useful team and a working delivery process.
Start a new venture around AI teams, or bring them into your existing SME or startup. We connect your tools, knowledge and workflows, starting with one process you can test.
Start with a scoped implementation. From €5,000.
AI-native means organizing work around connected AI capabilities, shared knowledge and clear human responsibility. We help you build that operating model progressively.
For founders creating a new organization: define the work, knowledge base, agent responsibilities and human decisions before adding more tools.
Begin with the first useful team and a working delivery process.
For SMEs and startups: map existing processes, connect the tools your people use and test the first workflow on real cases.
Expand into connected workflows when quality and value justify it.
Content, product, sales and operations are starting points. We select the smallest useful team and connect one workflow before expanding.
Examples from our own products and an anonymized client workflow. These show specific tasks; broader organization-wide results have not been measured.
The agent selects a topic from our knowledge and editorial plan, checks published and pending work, prepares the draft and sends it to the founder for review.
You keepNothing is published until the founder approves it.
Your team does not quote from the price list.
They quote from memory, and correct the price list in their head.
Everyone knows the day rate went up in March.
Everyone knows the travel surcharge only applies outside the city…
The agent reads the live pages, compares them with what the product actually ships, and prepares the correction in every language. After the release it checks the live pages again.
You keepThe founder merges every change.
After a visit, the investor supplies notes, photos or a plan. The agent organizes available dimensions, applies the investor’s unit costs and prepares a renovation estimate and return range. Missing measurements stay explicit.
You keepHe checks every estimate before making an offer.
Our approach, concrete workflows and first-hand examples. These resources are currently in English.
Build a new venture around AI teams or transition an existing SME or startup with shared knowledge, connected workflows and clear human ownership.
A first-hand example of an agent checking a website claim, preparing a correction in six languages and handing the change to a human reviewer.
Measure AI-team costs per accepted output, including retries, tools, hosting and human review. A practical framework before choosing models or adding roles.
Our public fork builds on Hermes Profile Packs with role-owned skills, defined handoffs and team recipes. The source is open for review and installation.
114
public profiles
Engineering, QA, product, marketing and operations specialists.
22
public profiles
Focused support for personal goals and decisions, with separate context.
30
public profiles
Teaching profiles for technical and professional learning.
8
public profiles
An expert panel for Belgian and French property analysis; not a substitute for regulated professionals.
Team Recipes select a minimal, recommended or expanded set of roles. Start small and add experts only when their work is needed.
Product, architecture, engineering, independent review and QA.
software-deliveryInvestment strategy, renovation costing and questions for legal and tax professionals.
property-deal-reviewPreview the exact profiles in a minimal team:
python install.py --recipe software-delivery \ --tier minimal --dry-runA recipe installs profile distributions; it does not automatically configure credentials, scheduling, memory or live execution.
Explore the open-source fork on GitHubIndependent fork of Dadmin88/hermes-profile-packs · AGPL-3.0 license · Built for Hermes Agent.
Begin with the smallest coherent set of specialists to avoid unnecessary delegation.
For managed setups, we define model choices, usage reporting and budget checkpoints before deployment.
Sensitive sends, publications, merges and other consequential actions have explicit approval boundaries.
We use Hermes Agent to configure specialist AI teams around your knowledge, tools and approval rules. We test one workflow on real cases before expanding. The first focused workflow is targeted in 14 days, subject to scope and access.
Map the work, name its owner and agree on inputs, acceptance criteria and a manual baseline.
Connect knowledge and tools, assign the smallest useful team and define approval boundaries.
Compare outputs, review exceptions and measure cost before putting the workflow into normal use.
Maintain the first workflow, review results and agree the next process separately.
The long-term goal is an AI-native organization. Each engagement has a defined scope, ownership and operating budget. Open-source profiles remain free.
An organization where AI teams are part of how work gets done: they use shared knowledge, follow defined processes and connect to business tools. People set goals, own decisions and control sensitive actions.
Fourteen days is a target for one focused first workflow, subject to scope, access and testing. A wider transition is delivered in stages, with each next step based on what the first workflows demonstrate.
We agree the boundaries for each task. Research, preparation and checks can run within those boundaries. Sending, publishing, merging, spending and other sensitive actions require the approvals defined for that workflow.
We review the cause, update the documented rules and add a regression case where appropriate. We recheck the workflow; a correction is not a guarantee that mistakes cannot recur.
We agree hosting, access and ownership before implementation. Open-source profiles are free; a managed engagement covers adaptation, integration and operation. We document provider accounts, running costs and how your team can take over.
Presetify is the AI-native organization and agent implementation service line of IES (Intech Engineering Solutions), led by Aurélien Gekiere. The Hermes profile-pack fork is independent of Nous Research.

You talk to the person who builds your agent, from the first call to the day it runs on your work.
Tell us whether you are building a new organization or improving an existing one, the first workflow you have in mind and the tools involved.