ChatGPT-maker’s CEO is promising to release an upgrade to ChatGPT that will bring its personality back to the old glory of ChatGPT-4o.
Sam Altman’s promise was triggered by widespread disappointment in the release of ChatGPT-5 earlier this year, which many believe gutted ChatGPT’s personality, making it feel cold and distant.
Observes Altman in an Oct. 14 tweet: “In a few weeks, we plan to put out a new version of ChatGPT that allows people to have a personality that behaves more like what people liked about 4o — we hope it will be better!
“If you want your ChatGPT to respond in a very human-like way, or use a ton of emoji, or act like a friend, ChatGPT should do it — but only if you want it.”
Should Altman’s promise play-out as hoped, it could be celebrated by ChatGPT fans as the best year-end gift they could receive this year.
In other news and analysis on AI writing:
*Google Offers Free Course on AI and Journalism: Journalists looking for the latest on how to level-up their AI skills may want to check-out a new, free, four-week course Google is offering on the topic.
Co-sponsored by the Knight Center for Journalism, the deep dive is designed to show journalists – and similar researchers and writers – how to leverage Google AI tools like Google Gemini, NotebookLM and Pinpoint in their day-to-day work.
Observes writer David Gewirtz: The course is aimed at journalists, “but if you’re a student or a writer, you could learn a lot, too.”
*ChatGPT Just Got a Privacy Upgrade: Thanks to some legal maneuvering, ChatGPT users looking to get their chats from the service easily deleted can now do so.
Previously, those chats had been forced to stay in preservation limbo – deleted from sight but still preserved by ChatGPT’s maker OpenAI – due to a court fight between OpenAI and The New York Times over alleged copyright infringement.
Fortunately for users, The New York Times has agreed to release OpenAI from that court-ordered preservation.
*Microsoft to Students in Washington State: C’mon and Take a Free Ride: Tech goliath Microsoft – which calls the state of Washington its home – just delivered lots of free access to AI for high school and college students there.
Under the giveaway, Washington high school students get three free years of Copilot Chat, Microsoft 365 desktop apps, Teams for Education, and Learning Accelerators – which are AIs designed to help students study.
Meanwhile, community college students in the state get 12 free months of Microsoft 365 Personal – Microsoft’s AI-powered productivity suite.
*Microsoft Beefs-Up AI in Windows 11: Microsoft is rolling-out a new feature of Windows 11 that enables you to access onboard AI with your voice.
Key features that will be accessible via voice include AI assistants, AI agents and AI contextual intelligence.
Observes Michael Nunez: “Starting this week, any Windows 11 user can enable the ‘Hey Copilot’ wake word with a single click, allowing them to summon Microsoft’s AI assistant by voice from anywhere in the operating system.”
One caveat: While AI agents have been sold hard as magical assistants that can complete multi-step tasks for you autonomously, in practice, those agents don’t work all the time.
*Discount Version of ChatGPT Rolling Out in Asia: ChatGPT’s maker OpenAI is rolling-out a bargain version of the chatbot across 18 Asian countries — part of its effort to stay the number one chatbot on the planet.
Priced at under $10/month, ‘ChatGPT Go’ is essentially a stripped-down version of ChatGPT Plus, a $20/month version that offers more than ChatGPT’s free version.
Observes writer Nina Raemont: “This subscription tier is available in 18 select countries, including India, Afghanistan, Bangladesh, Pakistan and Indonesia.”
*Salesforce Deepens Integrations with OpenAI and Anthropic: Salesforce – a sales, marketing and service suite designed for customer relationship management – just got tighter integration with key AI players OpenAI and Anthropic.
Observes Reuters: “The deals, announced on Tuesday, will embed OpenAI’s latest GPT-5 model and Anthropic’s Claude family of models directly into Salesforce’s ecosystem, enabling employees and consumers to interact with customer data and analytics in ChatGPT, Slack and Salesforce’s own software.
“The twin deals underscore Salesforce’s push to make its Agentforce 360 platform a central access point for major AI models, as enterprise software makers race to integrate generative AI tools into everyday business workflows.”
*Google Rolls Out Gemini Enterprise: Google has created a special version of its Gemini AI for business — Gemini Enterprise.
Based on the Gemini chatbot, which is Google’s answer to ChatGPT, Gemini Enterprise is designed to be the “front door” to AI for every employee at a business, according to Thomas Kurian, CEO, Google Cloud.
Observes Kurian: “This complete, AI-optimized stack is why nine of the top 10 AI labs and nearly every AI unicorn already use Google Cloud.”
*Oops, Sorry Australia, Here’s Your Money Back: Consulting firm Deloitte has agreed to refund the Australian government $440,000 for a study both agree was riddled by errors created by AI.
Observes writer Krishani Dhanji: “University of Sydney academic Dr. Christopher Rudge — who first highlighted the errors — said the report contained ‘hallucinations’ where AI models may fill in gaps, misinterpret data, or try to guess answers.”
Insult to injury: The near half-million-dollar payment is only a partial refund to what the Australian government actually paid for the flawed research.
*AI Big Picture: Despite AI Bubble Fears, Goldman Sachs Goes All In on AI: Global investment bank powerhouse Goldman Sachs is creating a special company team to go after the billions in AI infrastructure deals designed to build-out the new AI data centers, energy power plants and similar that AI titans insist the world will need for the coming AI boom.
Observes writer AnnaMaria Andriotis: “The effort is being fueled by the new wave of multibillion-dollar deals that involve financing artificial-intelligence data centers, the massive amounts of power they need to run, and the processing units behind the AI build-out.
”The new team will also focus on the building or upgrading of traditional infrastructure, ranging from toll roads to airports, in developed and emerging markets.”
Share a Link: Please consider sharing a link to https://RobotWritersAI.com from your blog, social media post, publication or emails. More links leading to RobotWritersAI.com helps everyone interested in AI-generated writing.
–Joe Dysart is editor of RobotWritersAI.com and a tech journalist with 20+ years experience. His work has appeared in 150+ publications, including The New York Times and the Financial Times of London.
The post ChatGPT Upgrade Promised By Year’s End appeared first on Robot Writers AI.
Claire chatted to Yuen Ting Chan from Natural History Museum about using robots to automate molecular biology experiments.
Yuen Ting Chan has nearly 20 years of experience working on translating, developing and optimising laboratory protocols, from DNA forensics to the biomedical field. She has brought automation to molecular laboratories for over 12 years, translating the laboratory protocols into bespoke scripts for a wide variety of liquid handling instruments. Her role at the Natural History Museum is to bring automation to the molecular laboratories, thus providing more opportunities for researchers to work on projects with large sample numbers for the wide variety of specimens within the museum.
Agentic AI is moving fast. In post one of this series, we looked at why agentic AI will fail without an AI gateway — the risks of cost sprawl, brittle workflows, and runaway complexity when there’s no unifying layer in place. In post two, we showed you how to tell whether a platform qualifies as a true AI gateway that brings abstraction, control, and agility together so enterprises can scale without breaking.
This post takes the next step, giving you a readiness check to avoid painful missteps or costly rework.
The risk is clear: The more progress you make without a gateway, the harder it becomes to retrofit one — and the more exposure you carry.
A true AI gateway needs to be customizable and future-proof by design, adapting as your architecture, policies, and budget evolve. The key is starting fast with a gateway that scales and adjusts with you rather than wasting effort on brittle builds that can’t keep up.
Let’s walk through the essential questions to help you assess where you stand and what it will take to support an AI gateway.
Before you decide whether you’re ready for an AI gateway, you need to know where your organization stands. Most AI leaders aren’t starting from zero, but aren’t exactly at the finish line, either.
Here’s a simple framework to pinpoint your AI maturity level:
Most enterprises today sit between Stage 2 and Stage 3 of their agentic AI journey. Pinpointing your current stage will help you determine what to focus on to reach the next level of maturity while protecting the progress already achieved.
Waiting until “later” is what gets teams in trouble. By the time you feel the pain of not having one, you may already be facing rework, compliance risk, or ballooning costs. Here’s how your readiness maps to the maturity curve:
Gateway thinking should begin toward the end of this stage when your infrastructure is ready and early experiments are underway. This is where you’ll want to start identifying the control, abstraction, and agility you’ll need as you scale, because without that early alignment, each new experiment adds complexity that becomes harder to untangle later. A gateway lens helps you design for growth instead of patching over gaps down the road.
This is the ideal window of opportunity. You’ve got one or two use cases in production, which means complexity and risk are about to ramp up as more teams adopt AI, integrations multiply, and governance demands increase. Use this stage to assess readiness and shape gateway requirements before chaos multiplies.
That means looking closely at how your pilots are performing, where handoffs break down, and which controls you’ll need as adoption spreads. It’s also the time to define baseline requirements, like policy enforcement, monitoring, and tool interoperability, so the gateway reflects real needs rather than guesswork.
Ideally, you should already have a gateway by this stage. Without one, you’re likely duplicating effort, losing visibility, or struggling to enforce policies consistently. Implementing governance without a gateway makes scaling difficult because every new use case adds another layer of manual oversight and inconsistent enforcement.
That opens hidden gaps in security and compliance as teams create their own workarounds or bypass approval steps, leaving you vulnerable to issues like untracked data access, audit failures, or even regulatory fines.
At this point, risks stop being theoretical and surface as operational bottlenecks, mounting liability, and roadblocks that prevent you from moving beyond controlled experimentation into enterprise-scale adoption.
It’s not too late for an AI gateway at this point, but you’re in the danger zone. Most workflows are live and the number of tools you’re using has multiplied, which means complexity and scale are increasing rapidly. A gateway can still help optimize cost and observability, but implementation will be harder, rework will be inevitable, and overhead will be higher because every policy, integration, and workflow has to be shoehorned into systems already in motion.
The real risk here is runaway inefficiency: The more you scale without central control, the more complexity turns from an asset into a burden.
This is the point where rolling out an AI gateway gets painful. Retrofitting at this stage means ripping out redundancies like duplicate data pipelines and overlapping automations, untangling a sprawl of disconnected tools that don’t talk to each other, and trying to enforce consistent policies across teams that have built their own rules for access, security, and approvals. Costs spike, and efficiency gains are slow as every fix requires unlearning and rebuilding what’s already in use.
At this level, not having a gateway becomes a systemic drag where AI is deeply embedded organization-wide, but hidden inefficiencies prevent it from reaching its full potential.
TL;DR: Stage 2 is the sweet spot for standing up an AI gateway, Stage 3 is the last safe window, Stage 4 is a scramble, and Stage 5 is a headache (and a liability).
Even if you’re early in your maturity journey, an AI gateway only delivers value if it’s set up on the right foundation. Think of it like building a highway: You can’t manage traffic at scale until the lanes are paved, the signals are working, and the on-ramps are in place.
Without the basics, adding a central control system just creates bottlenecks. So, if you’re missing the essentials, it’s too soon for a gateway. With the basics under your belt, the gateway becomes the load-bearing structure that keeps everything aligned, enforceable, and scalable.
At minimum, here’s what you should have in place before you’re ready for an AI gateway:
You don’t need dozens — just enough to prove AI is delivering real value. For example, your support team might use an AI assistant to triage tickets. Or finance could run a workflow that extracts data from invoices and reconciles it with purchase orders.
Why?: A gateway is about scaling and governing what already exists. Without real, active use cases, you have nothing to abstract or optimize. Think about the highway example above: If there’s no live traffic on the road, there’s nothing for signals to manage.
Your environment should already include some mix of:
Why?: A gateway is most effective when it can connect and coordinate across components. These are your lanes, signals, and interchanges. They may not be fancy, but they keep traffic moving. If your architecture is still theoretical, the gateway has nothing to route, secure, or govern.
A defined workflow should illustrate the path from raw input to real output, showing how your AI moves beyond theory into practice. It could be as simple as: LLM pulls from a vector DB → processes data → outputs results to a dashboard.
Why?: Gateways work best when they wrap around real flows — not isolated tools. Without at least one production workflow, you won’t yet have a demonstrated need for governance or observability for a critical system.
Regulations and internal mandates shape how AI should be designed, deployed, and monitored in your organization. From GDPR and HIPAA to enterprise audit requirements, these rules dictate data handling, access control, and accountability. An AI gateway becomes the natural enforcement point, embedding compliance and auditability into the workflow so that growth doesn’t come at the expense of security or trust.
Why?: Because the control layer of an AI gateway is what helps you meet those requirements at scale. These are your traffic laws and safety codes. As AI adoption expands, mandates multiply by use case, region, and department.
For example, a healthcare workflow may need HIPAA compliance, while a customer support bot handling EU data must follow GDPR. A gateway scales with that complexity, providing policy enforcement and auditability without manual effort.
A gateway can’t enforce what isn’t defined.
If your team hasn’t articulated what constraints the agentic AI needs to operate under, the success criteria it should meet, and the growth phases you defined, your gateway has nothing to optimize, secure, or scale.
A well-documented agentic AI strategy gives the gateway a clear mission and should spell out:
Every enterprise operates under mandates that define how AI is implemented and secured. The real question is whether your systems can enforce them automatically at scale.
An AI gateway makes at-scale enforcement possible. It embeds policy controls, access management, logging, and auditability into every agentic workflow, turning compliance from a manual burden into a continuous safeguard. Without that unified layer, enforcement breaks down and risks (including possible fines) multiply.
Consider the mandates your gateway needs to operationalize:
Governance and security are how you translate compliance intent into operational reality, and what keeps audit fire drills and access loopholes from derailing scale. Building on your regulatory mandates, your gateway should automate enforcement, consistently applying approvals, permissions, and audit trails across every workflow.
But your gateway can’t enforce rules you haven’t set. That means having:
A gateway doesn’t invent rules. It executes on the ones you’ve set. If you haven’t mapped who can do what — and under what conditions — you can’t scale agentic AI safely.
Every AI program reaches a point where cost control becomes strategy. A gateway helps you reach that point sooner, turning unpredictable, hidden costs into measurable efficiency gains. The setup investment pays itself back quickly once governance, observability, and scale are unified.
Without a gateway, costs are higher and harder to see: Teams lose time to manual reviews, DevOps hours pile up, and brittle architectures lock you into tools you’ve outgrown.
Multiply that across every use case, and missed savings compound into real financial strain.
A gateway eliminates those drains across several areas:
Think about two companies starting their agentic AI journey. Company A invests in a gateway early, while Company B tries to scale without it.
Company A’s return on investment (ROI) compounds over time. The upfront investment pays off through lower operating costs, faster innovation cycles, and reduced risk exposure. Company B may save upfront by skipping the setup costs, but the costs catch up later in rework, downtime, and missed growth opportunities.
Ultimately, the outcome is cost discipline that scales with your AI ecosystem — managing spend and turning compliance and agility into continuous ROI.
This readiness check is designed to help you avoid the missteps that slow AI maturity, from costly rework to mounting risk. The further you advance without an AI gateway, the more complicated it becomes to stand one up.
The best time to act is when early pilots start proving value. That’s the stage when oversight and scalability begin to intersect. By pinpointing where you sit on the maturity curve and confirming you have core use cases, foundational workflows, and clear policies in place, you can stand up a gateway that strengthens what’s already working instead of rebuilding later.
Whether you build or buy doesn’t matter. What matters is whether or not you’re prepared to support a gateway designed to match your architecture and enforce your policies while evolving with your budget.
If you’re ready to turn assessment into action, start with our Enterprise Guide to Agentic AI. It’s your roadmap for designing a gateway strategy that scales safely, efficiently, and without compromise.
The post How do you know if you’re ready to stand up an AI gateway? appeared first on DataRobot.