Building superglass: an AI Layer on Top of Salesforce, Not Another CRM

Five weeks. 408 commits. One principle we refused to break. superglass went from an empty repository to a working product between early May and mid-June, and almost every decision traces back to a single choice we made on day one: we were not going to build another CRM.

The first decision: read live, write back

Most tools in this space start by syncing your CRM into their own database. They copy your accounts, your contacts, your opportunities, and then they build features on the copy. The copy drifts. Now you have two sources of truth and a sync queue to babysit.

We decided not to hold any of your deal data. superglass reads your accounts, contacts, and activities live from the connected Salesforce org over OAuth. When a rep updates a stakeholder or moves a stage, we write it back to Salesforce. Your CRM stays the system of record. We are the intelligence layer on top of it.

This was the very first thing we built — Salesforce OAuth login and a live Accounts page, both shipped on day one. Everything after that inherited the constraint. It is harder to build this way. It is also the only honest way to be "an AI layer on top of Salesforce" instead of yet another place your data goes to rot.

One questionnaire, many documents

The heart of superglass is a questionnaire about how your team actually sells. Not MEDDIC-in-a-vacuum — your qualification criteria, your real pipeline stages, your buying committee, the moments you want managers to coach.

Answer it once and it becomes a live configuration object, with coherence rules that catch contradictions and a version history you can roll back. But the configuration is not the point. What it produces is.

From that one config, superglass generates your sales playbook, a manager coaching guide, rep onboarding sheets, a mutual action plan template, and a training deck. Change how you sell — adopt a new stage, add a committee role — and you regenerate all of it. The methodology stops being a deck that ages in a shared drive and becomes the thing that shapes how your CRM data is read.

Field Notes: from a voice memo to a Salesforce write-back

The feature we kept coming back to is the smallest one to describe and the hardest to build well.

A rep finishes a call and records a quick note — typed or spoken, in their own language. An AI pipeline reads it and pulls out the structured pieces: the evidence behind the deal, who was in the room, the buying signals worth acting on, and the ones worth worrying about. The rep reviews and confirms. Then superglass writes it back into the opportunity in Salesforce — activities, contact roles, the works.

No retyping into form fields after every call. No reconstructing the deal from memory when the manager asks "did you identify the economic buyer?" at the weekly review. The methodology gets applied at the moment of the work, not six months after the kick-off.

We also taught it to read the other direction: native Salesforce activity that already exists gets ingested as inferred evidence, so a deal that has never seen a superglass note still shows up with context.

The last 5%: a name and a mascot

For four weeks the product was called NeoCRM, then Glass CRM. Honestly, neither name said anything. In the final week we ran a proper naming exploration and landed on superglass — clear, a little playful, and it let us build a whole identity around the idea of a transparent layer you see your pipeline through.

Then we drew a mascot. Hungry is a small, hungry character who shows up on the homepage and in the product, and gives an early-access tool a face that isn't a stock illustration of a dashboard.

I'll admit I thought of branding as the cosmetic 5% you do at the end. It was more than that. The name forced us to articulate what the product is in one word, and the mascot gave the whole thing warmth that a feature list never will. The palette, the fonts, the domain cutover to superglass.io — all of it came in the last few days, and all of it changed how the product feels to use.

Where superglass fits

If your methodology lives in a deck and your deals live in Salesforce, those two things are probably not talking to each other. superglass is the layer in between — reading your real pipeline, capturing what happens in the field, and writing it back where your team already works.

We are still in early access. If this sounds like your setup, get started for free and tell us how your team sells today.