Inside a Notis onboarding with Kineon: from discovery call to compounding workflows

I had a really good onboarding and discovery call with Forrest Smith (Kineon) and Tyler Barrow (their automation lead), and it perfectly crystallized the kind of “first week” workflows that make Notis click.

The Notis paradigm: organize knowledge like a database, not like a folder

Most AI tools fail in the exact same place: retrieval. You can have a brilliant model and still get garbage outputs if the model can’t reliably find the right context at the right time.

The mental model behind Notis is pretty simple. Instead of treating your company’s knowledge as a pile of documents, you treat it as structured databases: meetings, tasks, customers, content ideas, ads, objections, proposals, onboarding notes. Each database has consistent fields, and each entry is a clean unit of context.

That structure is not “nice to have.” It’s how you make LLM retrieval sane. When you ask an agent to draft a sales follow-up, it doesn’t need “all notes.” It needs the deal, the last meeting, the objections, the current status, and the next step. Databases make that queryable.

Why Notis automations aren’t Make or Zapier (and why that matters)

Deterministic automation tools are incredible when the world is predictable. If A happens, do B, move C, post D. It’s brittle, but it’s fast and understandable.

Agent-based automation is for the messy part of work: interpreting intent, extracting decisions, turning a conversation into a plan, rewriting a hook in your voice, adapting an ad reference into something that fits your brand constraints. The output isn’t a single “correct” value. It’s a draft that needs judgment.

Notis is designed for that middle zone where automation is useful, but full autopilot is dangerous. That’s why we lean hard on human-in-the-loop validation. You let agents do the heavy lifting, then you approve, adjust, and ship.

Three high-leverage starting use cases for Kineon

What I liked about this conversation is that we didn’t try to boil the ocean. We focused on the workflows that compound.

Founder-led content engine: hooks and scripts to blog drafts

If you’re founder-led on content, your bottleneck is never “ideas.” It’s packaging. It’s turning messy insights into repeatable outputs without killing your calendar.

A good content engine looks like this: you capture raw thoughts in the format that’s easiest for you, usually voice or quick video. From there, an agent extracts hooks, proposes script options, and turns the winning direction into assets: a blog draft, a short post, a landing-page paragraph, whatever you need.

Notis is useful here because the content doesn’t live as a random transcript. It becomes entries with metadata: topic, audience, claim, proof, CTA, and the specific “voice constraints” that make it feel like you.

Ads and brand workflows: from inspiration to adapted creative and copy

The second workflow is ads, and it’s one of the most underrated places to use agents.

Teams collect ad inspiration constantly, but the gap is translation. An ad that worked for someone else is a reference, not a solution. You still need to map it to your product, your positioning, your claims, your proof points, your compliance constraints, and your brand voice.

With Notis, you can capture the reference, attach the why, and ask an agent to produce an adapted version that respects your Brand Kit and your existing best-performing angles. The output isn’t “post this.” It’s “here are drafts that are structurally similar to the reference, but true to your business.”

Calendar, meeting, and task capture once they move to Google

Kineon is moving from Microsoft to Google, and that’s where the third workflow becomes high leverage.

The fastest way to lose momentum is to let decisions die in calendars. Meetings happen, people nod, action items exist in everyone’s head, and then nothing makes it into a system that can be prioritized.

Once Google Calendar is the source of truth, you can build an agent workflow that captures meetings, drafts minutes, extracts decisions, creates tasks with clear owners, and keeps a rolling view of what matters this week. The trick is not creating more work. It’s turning “meeting exhaust” into structured entries you can actually query later.

Why I push teams to use Slack for support and debugging

One practical recommendation I gave them is to use Slack for support and debugging early on.

When you’re setting up agent-based automations, you want fast iteration. The worst thing you can do is burn tokens by repeatedly re-running a workflow just to understand why it behaved a certain way.

Slack makes that loop tighter. You can share the exact payload, the exact output, and the exact failure mode in one thread, and we can help you fix the prompt or the database structure without paying for ten extra runs. It sounds minor, but it’s the difference between “this is magic” and “this is expensive confusion.”

The webhook-driven meeting triage I personally use

To make it concrete, I shared the workflow I use all the time.

A meeting happens. I trigger a webhook that sends the transcript and context to Notis. Notis generates meeting minutes that are actually readable, then it extracts tasks, then it proposes follow-ups, and if the conversation contains a strong idea, it spins up a draft blog post.

The important detail is that each artifact lands in the right database entry. Minutes go to Meetings. Tasks go to Tasks. A draft goes to Blog. That’s the whole point of the database paradigm: the output isn’t just text, it’s text in the right place, connected to the right context.

Where Notis is at right now (and what I’m raising)

I’m still in that stage where product and distribution are the only two things that matter.

Notis has been bootstrapped so far. The plan is to raise 500k on a 4M pre to scale acquisition and build the kind of onboarding loops that make the product more self-serve without dumbing it down.

That’s the balancing act: keep the system powerful enough for real teams, while making it approachable enough that you don’t need an automation lead to get value.

Why Mexico might be a strong acquisition market (and the intros that matter)

A part of the conversation I didn’t expect, but really appreciated, was market validation.

Forrest offered to introduce me to well-educated, English-speaking local Mexican business owners. That’s exactly the type of group that can tell you, quickly, whether the pain is real, whether the messaging lands, and whether the product feels like leverage or like yet another tool.

He also mentioned potential introductions to China founder groups. Different market, different expectations, same core question: can we prove that Notis isn’t just impressive, but essential?

If we get this right, those conversations don’t just validate acquisition channels. They become the raw material for the best kind of proof: real stories, from real operators, explaining why the system changed how they work.

The punchline

This call reminded me that “AI automation” is not a category. There’s deterministic automation for moving data around, and there’s agent automation for turning messy reality into drafts you can approve.

The teams that win are the ones that treat context as an asset, structure it in databases, and build workflows where every meeting, every ad reference, and every idea turns into something shippable.