🔈 Listen on Spotify

Hi! Welcome to the 69th edition of the TomorrowToday newsletter.

We’re here to decode the AI chaos so you don't have to. Think of us as your friendly neighbourhood tech translators - we cut through the chaos, translate the jargon, and spotlight new AI tools that matter for founders, builders, and curious minds.

Buckle up, because the future's moving fast and we're here to make sure you don't get left behind! ⚡

If you enjoyed today’s newsletter, please forward it to a friend & subscribe by following this link.

_{~7 mins read}

/OpenAI /Agents /Personalisation

Remember Meta's Muse from edition 67? On 29 September, OpenAI joined the personal-agent race with dots: assistants with memory, connected tools and their own cloud computer that can keep working between conversations.

Give yours a name, talk to it naturally and hand over something that takes longer than one reply. The promise is an assistant that follows a project through, remembers the context and comes back when something needs your attention. Less explaining your life from scratch every time you open a chat.

It feels like the always-on agent idea becoming easier to use: fewer technical hurdles, more ordinary conversation. You can chat or call, then let it get on with the work. Whether it is genuinely helpful comes down to what it finishes.

The catch: access is rolling out on eligible plans, with a work allowance and approval boundaries. An always-on assistant still needs clear instructions about what it may do.

/Grok /Teams /Agents

We covered Grok's working bots in edition 63. On 28 September, Team Bots took the idea beyond one person's assistant: a shared bot with a role, useful tools and knowledge that a whole team can build on.

Think of an account-management bot that knows the handover process, or a marketing reviewer that understands the brand. Colleagues can work with it in Slack or Grok Bot, rather than each starting a separate assistant from scratch.

The useful distinction is shared expertise without automatically pooling everybody's private conversations. The documentation keeps individual chats separate, while team resources and permissions need deliberate setup. A useful colleague still needs the right keys.

The catch: Team Bots are in public beta on Teams and Enterprise. Before connecting anything, decide which files and credentials genuinely belong to the whole team. Start with one role, a narrow brief and somebody responsible for checking the results.

/OpenAI /Collaboration /Models

The rest of OpenAI's 29 September DevDay had a practical theme: give people and AI somewhere to work together. ChatGPT Space brings shared project knowledge and Pages into the workflow, so the brief and the work can live together.

Then there is GPT-6.1 Sol, the new model for coding and professional work. OpenAI describes performance approaching Astra's, at one-fifth of its standard input and output API prices. Sol costs $2 per million input tokens and $10 per million output tokens ( a bargain if you ask us).

That is OpenAI's comparison, not a promise that every task will cost five times less. The useful test is your own messy spreadsheet, proposal or codebase.

The catch: Space and Pages are on Pro, Business and Enterprise; collaborative slides were announced for the coming weeks. Sol launches in ChatGPT Work, Codex and the API, rather than ordinary Chat. Check the surface before hunting for the model.

In this section, we aim to spotlight an incredible AI tool or use case and guide you on how you can try it.

The fun part is choosing a name (JT calls his Watson because he thinks that he is Sherlock ;) ). The useful part is giving your new assistant a job that would otherwise sit on your list until Friday. Aim for one clear outcome: a comparison, a meeting brief or a tidy list of loose ends.

Start with OpenAI's setup guide and check that your account has access before changing your subscription. Your first dot is included on eligible Pro and Business Premium plans, with a gradual rollout. Plus does not include it; Enterprise access needs administrator enablement. South Africa is a supported country, although eligibility does not mean every account already has access.

Your first dot comes with a work allowance, rather than unlimited background work. Connecting an app also gives it access to information, so begin with something low-risk and understand the permissions before you accept. For a work account, follow your organisation's rules.

Don’t believe us? Try it yourself…

Pro tip: describe what ‘done’ looks like and where you want approval. ‘Find three options and show me’ is a much better first task than ‘sort out our next event’.

In this section, we're spotlighting real businesses using AI to solve actual problems.

Have you noticed how friendly the new AI agents are becoming? Muse has its little character. Grok Bot is presented as a teammate. OpenAI lets you name your dot and call it for a chat. These are systems built to take on serious work, yet meeting one increasingly feels like being introduced to a helpful, friendly and cute companion. But why?

We might find part of the answer in history. When electricity reached the White House in 1891, the Harrison family were afraid to touch the switches for fear of a shock. According to the electrician who installed the lights, he switched them on in the evening and returned to turn them off the next morning. The technology worked. Getting comfortable with it was another matter.

By 1926, electricity had acquired a friendly salesman: Reddy Kilowatt. Alabama Power's Ashton B. Collins created the cartoon character to humanise electric utility service, as the Smithsonian's archive records. More than 150 American utilities eventually licensed him. We cannot credit a cartoon with electrification, but the idea is easy to recognise: give something powerful and unfamiliar a personality people can welcome into their homes.

And there is a reason a friendly face can work so well. Big eyes, a rounded face and a large head are features researchers call ‘baby schema’. Experiments with infant faces found they increased perceived cuteness and the motivation to care. A study of robot faces found similar features could raise perceived trustworthiness too. Looking trustworthy is no guarantee of competence, but it can change the first impression before the machine has done anything.

Now come back to AI. Meta's designers explain that talking to a corporate logo felt odd, while an avatar felt more natural for an ongoing conversation. That is a revealing product problem. You can build an assistant capable of useful work and still leave people wondering how to approach it, what to ask or whether they want it around.

Our take? For plenty of everyday tasks, the models are already good enough to be useful. Part of the challenge now is helping people make room for them in their lives. A name, a voice and a friendly character make that first interaction feel less technical. This is how we read the shift, rather than proof that every AI company is following the same strategy.

For founders, that is worth paying attention to. Another capability matters when people actually use it. If your team has powerful tools sitting idle, consider how those tools are introduced and how easy the first useful task feels. Reddy Kilowatt gave electricity a familiar face. AI's new companions may be trying to make capable software feel equally at home.

AI is full of jargon, and we’re here to decode it. Each week, we’ll give you a plain-English definition of a buzzy term you’ve probably seen (but never fully understood).

Thanks for reading TomorrowToday! We’d love to hear from you:

➡️ What would you like us to cover next?
➡️ Have a tool or topic we should feature?

We’re building this with (and for) you. 🚀
See you next Tuesday 👋