Perplexity adds memory, publishing and private company data

Perplexity has quietly shipped four updates that change how the platform behaves day to day. On 14 July 2026, the company published a changelog covering self-improving memory, instant website publishing, access to private company financial data, and real-time switching between AI models mid-task. None of these are headline-grabbing announcements on their own. Together, they represent a meaningful shift in what a search-led AI platform can do for a small business.

Four features, one quieter platform shift

The centrepiece update is called Brain. It builds a private context graph from your past searches, uploaded files, and previous decisions. Each time you return to Perplexity, it draws on that graph to give more accurate, contextually relevant results. You no longer need to re-explain your business, your competitors, or your preferred report format every time you open a new session.

Before Brain, every Perplexity session started from scratch. That was a real time drain for anyone using the platform for recurring tasks. Brain removes that friction entirely.

The remaining three features each address a distinct gap. Website publishing lets you deploy a page to pplx.app or a custom domain in one click. Private company data gives you access to financial records via Forge Global, without a separate database subscription. Model orchestration lets Perplexity switch between models such as Claude and Opus mid-task, choosing the right one for each step automatically.

Why marketing and sales teams will notice this first

AI adoption is 80% people and process, 20% technology. These updates matter not because the features are sophisticated, but because they remove the steps that cause teams to abandon AI tools after the first few weeks.

For marketing professionals, the combination of Brain and one-click publishing is significant. A marketer running weekly competitor reviews no longer needs to re-enter brand context or preferred output formats. The platform remembers. A prototype landing page or internal dashboard can be published in seconds, bypassing a development queue that might otherwise take days.

For sales professionals, the private company data access is the headline feature. Background research on a prospective partner or client previously meant paying for a database subscription or spending an hour piecing together public records. That task is now handled inside the same platform where the sales team is already working.

The model orchestration feature matters for both roles. When Perplexity switches models mid-task to manage cost and output quality, it is doing something that most small businesses have not yet built into their own AI workflows. This is the direction the market is moving: away from single-model dependency and towards intelligent selection based on the task at hand. At gecco, our AI Assistants work is built around mapping exactly this kind of structured, vendor-neutral approach for UK SMEs.

Three actions to take this week

1. Enable Brain in your Perplexity account settings and upload two or three reference documents your team uses repeatedly, such as a competitor matrix, a brand brief, or a standard report template. This takes under 30 minutes and immediately reduces the context-setting overhead on every future session.

2. Ask your sales team to run one prospect background check using the Forge Global integration before your next pipeline review. Compare the time taken against your current process. You do not need a new subscription to do this. You need a Perplexity Pro account, which you may already have.

3. Identify one recurring marketing deliverable, such as a weekly summary page or a campaign brief, and test the one-click publishing feature to a pplx.app subdomain. This requires no development resource and can be live within minutes. Use it to demonstrate the concept internally before committing to a custom domain.

Considerations and limitations

Brain stores data privately within your Perplexity account, but it is worth reviewing what files you upload, particularly if your business handles commercially sensitive information. Perplexity is a US-based company, so data residency and processing locations should be checked against your own data handling policies before uploading client records or internal financial documents.

The Forge Global private company data integration is US-centric in its current form. Coverage of UK private companies may be limited, and you should verify the depth of records available for the specific businesses you want to research before relying on it for due diligence.

Model switching is automatic and not currently user-configurable. If your team has a preference for a specific model for compliance or consistency reasons, that preference may not always be honoured mid-task.

Where to go from here

If your marketing or sales team has been running the same AI searches on repeat, re-entering the same context week after week, that is a process gap these updates can close. Taking the AI Readiness survey gives you access to 65+ free resources and a custom AI Readiness report built around your business. That report is the starting point for a free 45-minute AI Readiness call to walk through the findings together.

Perplexity's model orchestration update is a useful signal about where assistant design is heading. At gecco, our AI Assistants work is built around exactly that shift: helping UK SMEs move from single-tool dependency to structured, vendor-neutral assistant workflows that select the right model for the right task.

Website · LinkedIn · Case Studies · Newsletter