devlin.ai

You describe who your learner is and who they need to practice talking to, and devlin.ai builds a conversational simulation with evaluation criteria you control. The result embeds in Articulate Storyline, Rise 360, or your LMS. You can also share the projects you create with our hosting. More than 1,700 instructional designers build on it today, from portfolio builders on the starter plan to enterprise learning teams planning large rollouts. Feel free to try it yourself:

It started as a workshop I couldn't scale

Since 2024 I've been running weekly AI project workshops in Peck Academy. I help people build AI projects and stay up to date with what hiring managers are asking for. I share takeaways from my own projects, and the attendees get feedback from each other.

In March 2026 I used one of these workshops to teach people how to add an AI-powered character to their Storyline course using a server, so it would be secure. Some students got there. Everyone else hit a different wall: Windows machines threw bugs the Macs didn't, and people came in with such different technical backgrounds that the same instructions landed completely differently.

Afterward I had two options:

• I could spend months producing a ten-part tutorial series on API keys, hosting, and security.
• I could build a tool that would handle all of that for them. The obvious winner.

I spent a week prototyping the tool and posted a video of it on LinkedIn, and the post blew up. A lot of the interest came from enterprise teams, which showed me there was real demand there. I knew that if this tool was going to serve the teams that needed it most, it had to survive enterprise security reviews. That's when I started working on SOC 2 compliance (way earlier than traditional advice suggests).

Forty-five seconds to a first draft

On demo calls I create a simulation based on one of the client's own use cases. I ask them who the learner is and who they need to practice talking to, and about 45 seconds later, the platform hands us a live simulation built exactly for them. It still needs refinement and testing after that, but clients are always surprised by how easy it is to get up and running and how convincingly the character plays the person they described. They also get excited about using it for their team.

Simulations, coaches, and hosting

Simulations are the core. Each one is anchored in a moment of performance (handle an objection, run a discovery call, interview a subject-matter expert), played against a character with realistic resistance, and finished with a defined end: the learner passes or fails against criteria that map back to the objectives. Designers build them in a guided builder or by talking to the built-in assistant, and learners respond by typing or by voice. Alongside simulations, designers can set up AI coaches that give learners ongoing guidance and feedback beyond a single scored scenario, and every project can live on devlin.ai hosting rather than a server the designer has to run.

I did not want to build this as an enterprise-only company. Most platforms doing AI roleplay are tailored to sales teams; I wanted one tailored to instructional design, with real access for individuals, and the plans reflect that. Portfolio builders and aspiring instructional designers use the starter plan to build simulations and host their projects on our infrastructure, with mostly the same security and benefits the enterprise teams get. One-stop shops use the core plan for client portfolios and smaller rollouts, while learner rollouts at scale belong on the bigger plans. It is a way to give back to the community that built my career while helping our field adopt this technology on terms tailored to us. It is also the natural extension of my years of scenario design work.

Security and scoring (the unglamorous parts)

The hard part has been enterprise security, and it wound up surprising me with how rewarding it became. We completed a SOC 2 Type 1 audit, run independent penetration tests, and keep adding to a security stack for ongoing monitoring and defense. Every new control has given me more confidence in the tool and more peace of mind as we plan large enterprise rollouts. The biggest win yet has been passing a Fortune 50 security review in a highly regulated industry.

The evaluation engine is where I've put the most effort so far. If companies are going to rely on these scores across tens of thousands of learners in a rollout, the scoring has to be fair, consistent, tied directly to their criteria, and free of bias. We regression-test it like the product depends on it, because it does.

Simulations also write results back into the course's own variables in Storyline, so pass or fail, scores, and completion become part of the course logic designers already use. Between the conversational data and the scoring, managers and leaders get visibility that goes way beyond SCORM completion data.

Results

The best part about this project is the opportunities we've been getting to impact real business goals. One team came to us because their revenue-generating employees were losing hours to role-playing with trainees. The simulations and AI-powered feedback can now carry that load. Trainees practice on their own time in a consistent format. The team can get scores and transcripts for every session, and the trainers can get back to the work that pays for everything else.

Where this goes

The ambition is for devlin.ai to be the platform for practice in our field. Designers create learning journeys, learners work through them, and the platform shows each learner's progress, performance across the organization, and how ready the workforce actually is, measured through conversation instead of course completions. Because every session generates data about what effective practice looks like, the system can keep improving at helping learners get better at their jobs. This is my long-term goal: a smart, self-improving practice layer for the industry.

The same simulations are already the practice layer inside Peck Academy's curriculum, where students run simulated subject-matter-expert interviews and stakeholder conversations before they face real ones. And it is the same team and the same AI-first workflow that built DevlinOS; each system makes the other better.

If you want to see the product itself, visit devlin.ai.